« au service de l’analyse » — since 1998
vol. 20 I n°2 I 2018
Regulation for Artificial
Intelligence and Robotics in
Transportation, Logistics and
Supply Chain Management
Network Industries Quarterly | Published four times a year, contains information about postal, telecommunications, energy, water, transportation and network industries in gen-eral. It provides original analysis, information and opinions on current issues. The editor establishes caps, headings, sub-headings, introductory abstract and inserts in articles. He also edits the articles. Opinions are the sole responsibility of the author(s).
Subscription | The subscription is free. Please do register at fsr.transport@eui.eu to be alerted upon publication.
Letters | We do publish letters from readers. Please include a full postal address and a reference to the article under discussion. The letter will be published along with the name of the author and country of residence. Send your letter (maximum 450 words) to the editor-in-chief. Letters may be edited.
Publication director | Matthias Finger
Managing editor | Irina Lapenkova, David Kupfer Publishing editor | Ozan Barış Süt
Founding editor | Matthias Finger Cover image | Stock photo ID:852015986
Publisher | Chair MIR, Matthias Finger, director, EPFL-CDM, Building Odyssea, Station 5, CH-1015 Lausanne, Switzerland (phone: +41.21.693.00.02; fax: +41.21.693. 00.80; email: mir@epfl.ch; website: http://mir.epfl.ch/)
Network Industries Quarterly, Vol. 20, issue 2, 2018 (June)
Regulation for Artificial Intelligence and
Robotics in Transportation, Logistics and Supply
Chain Management
Editorial Introduction by the Guest Editor
Under the terms of Internet of Things, Industry 4.0 and Physical
Internet as well as several others, many automatization and
digitalization trends are on the move for the transportation, logistics and supply chain sector. Many technology aspects are driving these developments, in line with economic aspects. But increasingly also questions of human perception, motivation and safety are entering the discussion, emerging as a crucial topical area for overall economic impact and success.
Regulation for technology developments in artificial intelligence and robotics are commonly seen as one of the important yet structurally neglected fields regarding the human perspective on increasing automatization. This was highlighted in 2017 by the European Parliament report and a public consultation, indicating that a vast majority of citizens in Europe is regarding those developments as positive innovation fields but where further safeguards and regulations are needed, see the EP Resolution on Civil Law Rules on Robotics, 2015/2103(INL).
This issue is connected to an innovation workshop that took place on February 26 2018 at the Florence School of Regulation and directed at discussing the state of the art within the field of transportation, logistics and supply chain management. Furthermore, an evaluation regarding possible actions like regulation, agency- or industry-based approaches for establishing safeguards towards effective but risk-mitigating settings for this sector is aimed for.
Initial contributions collected here are directed at providing an interdisciplinary overview regarding the perspectives of industry and logistics actors, researchers in the economic, computer sciences, law and sociology domains as well as other interested parties from the field of political actors and associations. This shall enable the start of an open discussion what sorts of regulation are necessary in order to secure human trust and motivation in AI and robotics developments without placing too much of a burden to the economic development in the transportation, logistics and supply chain sector.
3
Regulation for Artificial Intelligence and
Robotics in Transportation, Logistics, and
Supply Chain Management: Background and
Developments
Matthias Klumpp, Caroline Ruiner
8
Regulation for Artificial Intelligence and
Robotics in Transportation, Supply Chain
Management, and Logistics
Julian Sanders
11
How to prepare workers for logistics
innovations today and tomorrow
Dominic Loske
16
The impact of digital technologies and
artificial intelligence on production systems
in today Industry 4.0 environment
Francesco Pilati, Alberto Regattieri
21
Liability and automation: legal issues in
au-tonomous cars
Giuseppe Contissa, Francesca Lagioia, Giovanni Sartor 27
Announcements
dossier
contents
The guest editor of this special issue is Dr. Matthias Klumpp (Diploma University of Leipzig, M.A. University of Kassel, PhD University of Leipzig). Matthias Klumpp is currently a professor for logistics at FOM University of Applied Sciences Essen, Germany. He is also senior researcher at University of Duisburg-Essen and Fraunhofer Institute for Material Flow and Logistics Dortmund, Germany. His most recent published work is addressing questions of digitalization, human resources and sustainability in transportation and logistics; it appears in
International Journal of Logistics Research and Application and
Guest editor: Matthias Klumpp, Professor for Logistics at FOM University of Applied Sciences Essen and 2018 Visiting Fellow at the European University Institute
dossier
Regulation for Artificial Intelligence and Robotics in
Transpor-tation, Logistics, and Supply Chain Management: Background
and Developments
Matthias Klumpp*, Caroline Ruiner**
Digital solutions and artificial intelligence applications provide innovation potential for transportation, logistics, and supply chain management. However, the question of competencies, motivation, and acceptance of the human workforce is important for the practical success of such initiatives, in firms and in comprehensive transportation systems and networks. This section addresses the background for an inquiry into the framework conditions, recent developments and necessities for regulation of robotics and artificial intelligence in this field.
Background
T
here are a multitude of digitization and automa-tion developments in the transportaautoma-tion, logistics and supply chain management domain, high-lighted by concepts such as Internet of Things, Industry 4.0, or Physical Internet (Zhong et al., 2017; Fawcett & Waller, 2014). Technological aspects are the main drivers for these developments, and in most cases, they are aligned with economic factors such as cost savings or increasing customer reaction and time to market speed (Masoud & Jayakrishnan, 2017; Wojtusiak, Warden & Herzog, 2012). But besides these technical and economic issues, questions of competencies, motivation, and acceptance with the hu-man workforce are also increasingly entering the discussion and emerging as a crucial topical area for overall econom-ic impact and success (Mavrovouniotis, Li & Yang, 2017; Zijm & Klumpp, 2016). In this context, regulation inrobotics and artificial intelligence (AI) is commonly seen as an important yet underrepresented field related to the human perspective on increasing automation. This point was highlighted in 2017 by a European Parliament report
and a public consultation (Delvaux, 2017), which indicat-ed that many citizens in Europe regard developments in robotics and AI as positive innovation fields, but require further discussions and regulations (Table 1).
It can be recognized that in an overall perspective, AI as a future development trend is seen more critical than the use of robots – who are in many cases perceived as support and help to humans. In detail, this is connected to a majority of 83% of all respondents agreeing or strongly agreeing to the statement that robots are good for society as they help people – whereas only 34% of respondents agree or strongly agree to the statement that robots steal peoples’ jobs. Still, a huge majority of 92% also agrees or strongly agrees that robots are a technology that requires careful management, i.e. regulation and oversight. On the other hand, half of all respondents (52%) agree or strongly agree towards the statement that AI is a threat to privacy.
However, favoring AI application is the fact that only 26% of respondents agree or strongly agree to the statement that AI is a threat to fundamental human rights. Altogether, these statements represent the mixed perception of citizens
Strongly agree Agree Neither agree/ disagree Disagree Strongly
disagree Don’t know
Robotics
Technology requiring careful management 57 % 35 % 4 % 2 % 2 %
-Necessary for hard or dangerous jobs 57 % 35 % 5 % 2 % 1 %
-Efficient way for transport/delivery 37 % 34 % 16 % 9 % 3 % 1 %
Good for society as they help people 34 % 49 % 11 % 3 % 3 %
-Steal peoples’ jobs 9 % 25 % 31 % 25 % 9 % 1 %
Create inequity 5 % 13 % 21 % 32 % 24 % 5 %
Artificial Intelligence (AI)
Threat to privacy 20 % 32 % 22 % 14 % 8 % 4 %
Threat to humanity 13 % 16 % 23 % 22 % 22 % 4 %
Threat to fundamental human rights 12 % 14 % 25 % 22 % 20 % 7 %
Table 1. Public Perception on Robotics and AI Applications in Europe (Question: Please indicate if you agree or disagree with the following statements).
dossier
towards automation (robots) and digitization (AI) trends – hinting at a positive attitude of citizens towards a political regulation role in these fields. Regarding regulation fields, the report outlined the following six key areas of regulatory action in detail regarding robotics and AI application in the European Union, namely (European Parliament 2017, p. 8):
(1) Rules on ethics (2) Liability rules
(3) Connectivity, intellectual property, and flow of data (4) Standardization, safety, and security
(5) Education and employment
(6) Institutional coordination and oversight.
In order to discuss such regulatory action further, it can be distinguished between regulatory approaches, arguments for regulation and areas of regulation. Approaches can be divided into law regulation, agency-based regulation or industry-based approaches for establishing safeguards towards effective but risk-mitigating settings. Relevant impacts from the public consultation indicate significant public support for political regulatory action in this field due to the reports of citizen opinions and anxieties (Table 2).
As Table 2 shows, arguments for regulation are headed
by the items ‘data protection’ (85% of respondents are concerned or strongly concerned) and ‘values and principles’ (81% are concerned or strongly concerned). In addition, liability rules are an important argument in the eyes of citizens with 74% being concerned or strongly concerned about this issue. Smaller shares of the respondent group are listing arguments such as EU competitiveness (66% are concerned or strongly concerned), physical safety
(54% are concerned or strongly concerned), or intellectual property (44% are concerned or strongly concerned). Regarding areas of regulation, transportation is present very prominently in the top five areas with autonomous vehicles being number one (87% regard it as important or very important) and drones being number four (73% deeming regulation in this area important or very important). In-between medical and care robots are seen as necessary area of regulative action (with 80% and 73% deeming these areas to be important or very important respectively). In addition to this, the world of work has to be recognized too since this is where robotics and AI applications are implemented and people encounter them actively in cooperation.
Human Work and Digitization
The EU study presented indicates how citizens perceive robotics and AI applications, their anxieties and highlights certain approaches for the regulation of new technologies. The citizen’s perceptions are likely to match the perceptions of the workforce. Likewise do the key areas of regulatory action take into account the human factor in automated and digitized work settings which is important with regard to the employer’s due diligence obligations. Moreover, the human factor is of crucial relevance since workers’ perceptions affect the acceptance and their handling of robotics and AI (Ventakesh & Davis, 2000), and is, thus, central for the economic impact and
success of these new forms of work organization. A positive attitude towards work settings usually comes along with intrinsic motivation – defined as behavior coming from within an individual, out of interest for the activity and enjoyment – leading to high job performance and satisfaction, commitment and innovation (Deci & Ryan, 2000). Accordingly, if digitization in the working context is perceived as positive and supportive, it will promote
Strongly
concerned Concerned Neutral Not concer-ned at all Don’t know
Arguments for Regulation
Data protection 51 % 34 % 8 % 6 % 1 %
Values and principles 51 % 30 % 9 % 10 %
-Liability rules 35 % 39 % 19 % 6 % 1 %
EU competitiveness 29 % 37 % 22 % 8 % 4 %
Physical safety 26 % 38 % 22 % 11 % 3 %
Intellectual property 17 % 27 % 27 % 24 % 5 %
Very
important Important Neutral Somewhat important importantNot at all Don’t know
Areas for EU Regulatory Action
Autonomous vehicles 55 % 32 % 5 % 4 % 4 %
-Medical robots 48 % 32 % 12 % 5 % 3 %
-Care robots 38 % 35 % 15 % 8 % 4 %
-Drones 42 % 31 % 12 % 6 % 3 % 1 %
Human repair 40 % 32 % 13 % 4 % 2 % 4 %
Table 2. Public Expectations regarding Regulation Motivation and Areas in Europe. Source: Evas (2017), p. 11-12 (n=259)
dossier
pro-organizational behavior. In this respect, the careful management and design of digitization in the working context plays a key role and is the central leverage to orchestrate the workforce. Measures of Human Resource Management (HRM) therefore need to consider the use of robotics and AI in the work process based on an overall strategy implying specifications and rules, e.g. in terms of ethics, safety and security issues. Moreover, an adequate information of and communication with the workforce will promote liability and trust (Lewicki et al., 1998). For a positive attitude and efficient use of new technologies in work processes, it is crucial that workers perceive the support of the management, that the organization values their contributions and cares about their well-being (Eisenberger et al., 1986). To foster intrinsic motivation in digitalized work contexts, workers attach great importance to experiencing autonomy and their tasks as meaningful (Hackman & Oldham, 1980). This would also mean that the interaction between robots, AI and workers is seen as equally and free, meaning that workers perceive locus of control (Rotter, 1966). Therefore, it is essential that workers obtain the qualifications required for working with new technologies. Consequently, the alignment of the workers’ competencies towards changed work requirements is a basic prerequisite for the acceptance of robots and AI in work contexts. The acceptance of new technologies in work settings will also be supported when the process of automation and digitization is not only organized top-down but also bottom-up, i.e. when workers have the opportunity to participate in organizing and designing digitization processes, contributing their ideas and perspectives (Boxall & Purcell, 2003). Finally, the use of new technologies in organizations is not only designed in cooperation with management and workers but also involving actors like employee representatives (e.g. works council), securing the workers interests in data protection as well as general values and principles. This would eventually help to promote the understanding of robots and AI supporting the workforce and establishing a regulatory framework securing workers’ positive attitudes. Connected to these general HR concepts, one central motivation for regulatory action is to promote the worker’s acceptance and to mitigate possible human resistance towards robots and AI applications in transportation, logistics and supply chain management. Therefore, it is helpful to understand the structure of typical workers’ resistance towards automation within the field of transportation, logistics and supply chain management – this is depicted in the following Figure 1.
In many business application contexts three major resistance hurdles can be identified before a full human cooperation mode can be reached. First, workers have to accept single automation steps as AI competence, e.g. the competence of an automated steering system to handle truck cruise control orders. Second, it is even harder and usually met by a higher level or hurdle of human resistance
to accept independent AI decisions, e.g. suggestions by a navigation system in driving trucks. There might for example be higher rates of neglect, meaning navigation
suggestions are overturned in practice. Third, human actors have to accept AI autonomy, for example an autonomous steering system for trucks. In this case, the resistance might be highest as autonomous behaviour of automated systems brings about the highest level of fear and insecurity among human coworkers. In this area, regulation might therefore be needed the most and provide the most benefit: Regulation may help to reduce the volume and impact of these human resistance hurdles for an efficient human-computer interaction (HCI). This can be achieved as human workers may be able and motivated to start HCI settings with a lowered resistance of they know that regulations are in place safeguarding their physical safety and their personal data protection and employment rights.
Areas of Regulation in Transportation, Logistics and Supply Chain Management
Applying the six defined key areas of regulation as outlined in the EU study specifically towards the transportation, logistics and supply chain sector, the following observations can be derived:
(1) Rules on ethics: Especially in transportation – as public traffic is concerned – ethical rules of engagement are important, e.g. if accidents occur and split-second decisions have to be taken by automated systems like which deviation route to take with specific casualty impacts. The major problem in this area is, that such decisions have to be implemented beforehand within the automated and AI transportation systems, as in many cases reaction times will be too short for any human driver to contemplate and interfere with the autonomous steering e.g. of trucks and
Figure 1. Motivation for Regulation in Robotics and AI Applications in Logistics.
dossier
cars.
(2) Liability rules: Again, transportation as far as public transportation is affected will be a major development field for liability rules and within their wake insurance markets and products for automation and digitization. But also in the production logistics environment many important liability questions will arise, e.g. who might be liable for incorrect order volumes (order volume too high with subsequent warehousing costs or order volume too high with resulting production interruptions and market costs from customer contracts).
(3) Connectivity, intellectual property, and flow of data: Connectivity is a major concern for the transportation, logistics and supply chain management field as a global sector. Therefore, many research endeavors already explore the use of standardized industrial data spaces also for transportation. This will be increasingly important as many applications (like with smartphones) will arise for transportation and logistics settings, requiring a unified communication and interaction framework.
(4) Standardization, safety, and security: As transportation always includes a physical component, safety and security issues are highly important, affecting many public hubs (ports, stations and airports) as well as main lines throughout the countries and around the globe. Security issues might easily clog up passenger and cargo traffic, resulting in large economic losses as well as private burdens in terms of lost time and increased stress. Therefore, automation and digitization developments are urged to enhance the overall safety and security level in transportation, as well as providing this increase at lower economic and societal cost. (5) Education and employment: Work and qualification issues are very relevant in the logistics sector as it represents a personal-intensive service industry. Digitization is seen as an ambivalent trend regarding this as there are at the same time effects of eased work burden and facilitated training and education by electronic means as well as increased work burden and stress by the way of increased transparency and oversight or even job losses in specific areas – though it has to be emphasized that the total number of jobs is not expected to be reduced in the transportation sector for a long time to come. But it cannot be neglected that qualification requirements will change and therefore the importance of education and training, requiring also structuring and evaluation regulatory action from the authorities in this field.
(6) Institutional coordination and oversight: The interaction of different institutions in supply chains and global transportation will change, as on the one hand digitization and the use of AI will facilitate many processes and services along the transport ways. For example, document
translation can be automated in the near future, lower cost and time requirements in customs, transportation and logistics. This will on the other hand also require coordination among supply chain partners, as they have to agree on standards and cost sharing regimes for automated services.
An interdisciplinary perspective from different science disciplines is helpful in implementing such regulatory areas. This includes the perspectives of industry and logistics actors, researchers in the economic, computer sciences, law, and sociology domains, as well as other relevant parties from the field of political actors and associations. This could be an invitation to start an open discussion about what sorts of regulation are necessary in order to secure human trust and motivation in robotics and AI developments without placing too much of a burden on the economic development in the transportation, logistics, and supply chain management sector (Klumpp, 2018; Petit, 2017; Fors, Kircher & Ahlström, 2015).
Contributions and Outlook
The contributions of this issue are aligned with a multi-perspective analysis regarding the question of regulation for robotics and AI applications in transport, logistics, and supply chain management, intending to provide a sort of mapping of future research topics in this (Wieland, Handfield & Durach, 2016). At the same time, they are addressing different aspects from the described six key areas of regulation of robots and AI: The first two contributions start with a business practice perspective. Julian Sanders emphasizes the dynamic innovation requirements for logistics service providers on a global scale, hinting at the necessity of regulation in the area of ‘institutional coordination and oversight’. While following that, Dominic Loske outlines the challenges of a day-to-day transportation and logistics situation in urban food retailing, focusing on the question of human training for truck drivers facing ever-faster digital innovation steps; this is discussing the ‘education and employment’ area of regulative action as outlined above. From a production logistics research perspective, Francesco Pilati and Alberto Regattieri provide an outlook on what big data analysis can do for an improved ergonomics situation in production. This again is addressing the regulation areas ‘standardization, safety and security’ as well as ‘institutional coordination and oversight’, including the role of unions and other work safety institutions in the production logistics context. Giuseppe Contissa, Francesca Lagioia and Giovanni Sartor analyse the impact of automation in the allocation of liability within autonomous cars. They discuss the tasks allocation between human and automation, and the resulting responsibilities.
dossier
Altogether, the issue is aimed at sparking a discussion re-garding future questions for regulation towards the appli-cation of robotics and AI in the transportation, logistics and supply chain sector in order to allow for a smooth and efficient changeover with digitization trends. As can be recognized from this issue, there are manifold open gaps and research items to be applied in this context in the up-coming years.
References
Boxall, P., and Purcell, J. (2003). Strategy and Human Re-source Management. London: Palgrave Macmillan.
Delvaux, M. (2017). Report with Recommendations to the Commission on Civil Law Rules on Robotics. No. (2015/2103(INL)), Committee on Legal Affairs, Rapor-teurs for the Opinions: Georg Mayer, Committee on Transport and Tourism, Michał Boni, Committee on Civil Liberties, Justice and Home Affairs, Brussels.
Eisenberger, R., Huntington, R., Hutchison, S., and Sowa, D. (1986). ‘Perceived organizational support’, Jour-nal of Applied Psychology, 71 (3), 500–7.
European Parliament (2017). Consultation Document – Civil Law Rules on Robotics, Brussels: European Parliament.
Evas, T. (2017). Public Consultation on Robotics and Arti-ficial Intelligence – First Results of Public Consultation. Eu-ropean Parliamentary Research Service, Brussels. Available at: www.epthinktank.eu.
Fawcett, S.E., and Waller, M.A. (2014). ‘Supply chain game changers – Mega, nano, and virtual trends – And forces that impede supply chain design’, Journal of Business Logistics, 35 (3), 157–64.
Fors, C., Kircher, K., and Ahlström, C. (2015). ‘Inter-face design of eco-driving support systems – Truck drivers’ preferences and behavioural compliance’, Transportation Research Part C, 58 (D), 706–20.
Hackman, J.R., and Oldham, G.R. (1976). ‘Motivation through the design of work: test of a theory’, Organiza-tional Behavior and Human Performance, 16 (2), 250–79.
Klumpp, M. (2018). ‘Automation and artificial intelli-gence in business logistics systems: human reactions and collaboration requirements’, International Journal of Logis-tics Research and Applications, 21 (2), 224–42.
Lewicki, R.J., McAllister, D.J., and Bies, R.J. (1998). ‚Trust and distrust: New relationships and realities’, Acad-emy of Management Review, 23 (3), 438–58.
Masoud, N., and Jayakrishnan, R. (2017). ‘A real-time algorithm to solve the peer-to-peer ride-matching problem in a flexible ridesharing system’, Transportation Research Part B, 106 (1), 218–36.
Mavrovouniotis, M., Li, C., and Yang, S. (2017). ‘A sur-vey of swarm intelligence for dynamic optimization: Algo-rithms and applications’, Swarm and Evolutionary Compu-tation 33 (1), 1–17.
Petit, N. (2017). Law and Regulation of Artificial Intelli-gence and Robots – Conceptual Framework and Normative Implications. SSRN Working Paper, available at: https:// papers.ssrn.com/sol3/papers.cfm?abstract_id=2931339.
Rotter, J.B. (1966). ‘Generalized expectancies for inter-nal versus exterinter-nal control of reinforcement’, Psychological Monographs, 33 (1), 300–3.
Ryan, R.M., and Deci, E.L. (2000). ‘Self-determination theory and the facilitation of intrinsic motivation, social development and well-being’, American Psychologist, 55 (1), 68–78.
Venkatesh, V., and Davis, F. D. (2000). ‘A theoretical ex-tension of the technology acceptance model: four longitud-inal field studies’, Management Science, 46 (2), 186–204.
Wieland, A., Handfield, R.B., and Durach, C.F. (2016). ‘Mapping the landscape of future research themes in sup-ply chain management’, Journal of Business Logistics, 37 (3), 205–12.
Wojtusiak, J., Warden, T., and Herzog, O. (2012). ‘Ma-chine learning in agent-based stochastic simulation: Infer-ential theory and evaluation in transportation logistics’, Computers and Mathematics with Applications, 64 (12), 3658–65.
Zhong, R.Y., Xu, C., Chen, C., and Huang, G.Q. (2017). ‘Big data analytics for Physical internet-based intelligent manufacturing shop floors’, International Journal of Pro-duction Research, 55 (9), 2610–21.
Zijm, H., and Klumpp, M. (2016). ‘Future Logistics: What to Expect, How to Adapt’ in M. Freitag, et al., Dynamics in Logistics, Lecture Notes in Logistics (Cham: Springer International Publishing), 365–79.
dossier
Regulation for Artificial Intelligence and Robotics in Transportation,
Supply Chain Management, and Logistics
Julian Sanders*
Previous production-oriented developments in technical infrastructure and IT form the basis for recent trends of automatization and digitalization in the transportation, supply chain, and logistics sector. Along with economic aspects, areas such as human perception, motivation, and safety are gaining in significance against the background of artificial intelligence and robotics. Long seen as contradiction, these factors are now understood as a crucial enabler for overall economic impact and success. In particular, for the people-oriented logistics business, but also for many other industry branches, the question of regulation becomes increasingly important in order to secure human trust and motivation in artificial intelligence and robotics without raising the burden to economic development.
Introduction
T
hrough terms such as the Internet of Things, In-dustry 4.0, and Physical Internet, many automa-tization and digitalization trends are developing for the transportation, supply chain, and logistics sector. Both trends result largely from production-based develop-ments over the last years that have changed the role of data comprehensively (Figure1). Starting with the implemen-tation and operation of LVS systems in the 1970s, data became rapidly enabler for processes in the 1990s and also for products until the 2000s, and are today products them-selves. The different data development stages should not beunderstood as disjunct, rather as parallel developments in companies. For one thing, data are the result of digitaliza-tion and automatizadigitaliza-tion, but also a resource for service cre-ation or even products, which leads to the ‘data paradox’.
Considering that the number of connected devices has increased by a factor of almost 35,000 since the first con-nected devices were launched in 1992, the development of the Internet of Things has increased the number of connect-ed devices almost exponentially. By 2020 the total amount
of connected devices will double to 50 billion due to fur-ther development of the Internet of Things (Figure 2).
In summary, past and present trends of the Internet of Things, Industry 4.0, and Physical Internet are results of the development and understanding of data and its impor-tance. Therefore, the topical framework of automatization and robotics, enabled and driven by these developments, became one of the most important action points for a wide range of industries in the future. The application of au-tomatization and robotics, combined with artificial intel-ligence will be a major innovation in transportation and supply chain management, but also growth and efficiency
driver in the next 15 years. Focusing on challenges and opportunities for the logistics sector, the recent trends can even be described as a ‘game-changer’ for many different areas.
Risks and opportunities
The main exogenous drivers for the recent developments in digitalization and automatization are cost pressure (46 percent), staff shortage (64 percent), complexity (39 percent) and dynamically changing buying behavior (57
* Julian Sanders, M.A. (UCAM Universidad Católica San Antonio de Murcia, Spain), julian.sanders@me.com
Figure 1. Role of data in the course of digitalization and automatization Source: Otto, B. (2016)
dossier
percent), combined with endogenous factors stated as the need for transparency in supply chains (55 percent), busi-ness analytics (62 percent) and collaboration (55 percent). Next to the challenges, logistics companies consider op-portunities in terms of additional revenues (34 percent) and cost reduction (34 percent), (Kersten et al. 2017, p. 19). Many developments are driving these aspects with
en-abling technologies, also in line with economic success. In direct comparison, the opportunities of digitalization and automatization outweigh the counteracting risks of digital transformation (Figure 3).
Against this background, it can be noted that the vast majority of companies in transportation, supply chain management, and logistics expect fewer risks than oppor-tunities from digital transformation driven by digitaliza-tion and automatizadigitaliza-tion. Nevertheless, the logistics sector attaches high importance (59 percent) to the burden re-sulting from regulation and compliance related to digital transformation, making this a crucial topical area for over-all economic impact and success.
Regulation conclusions
Regulation for technology developments in artificial intel-ligence and robotics can be seen as an important yet struc-turally neglected field regarding the human perspective on increasing automatization. This point was underlined in 2017 by a European Parliament report and a public con-sultation for the European Union, which indicated that a majority of citizens in Europe regard those developments as positive innovation fields, but with the need for further
safeguards and regulations (see the European Parliament Resolution on Civil Law Rules on Robotics, 2015/2103 (INL)). Therefore, questions of human perception, moti-vation, and safety are increasingly entering the discussion (Ruiner & Klumpp, 2018a).
It has been found that human-AI collaboration decisively depends on its design requiring opportunities for
auton-omous decisions, feedback, and participation, as well as individualized and respectful communication of support and care. For acceptance and human-AI team-building, as well as proactive use of AI, digital devices and automa-tized robots must be designed to support humans, not to control or direct them. Thus, the preparation and partici-pation of the human workforce in combination with such applications as human-computer interaction (HCI) is an important issue for individuals regarding the acceptance and use of AI, for unions, politics, and regulation, as well as businesses, in order to retain competitiveness and design positive impacts. Furthermore, regulation issues require local, regional, national, and European actors to address standardization, safety, and trust issues for the public, as well as the European workforce in transportation and logistics. All societal groups and representatives are called on to help with this major task, as the world of work as well as general mobility will change significantly and will require structured guidance (Ruiner & Klumpp, 2018b).
As the aim of this article was to describe past, recent and future trends in digitalization and automatization against the background of regulation focusing on risks and oppor-tunities of the digital transformation, it should be noted
Figure 2. Development of connected devices Source: Mesh-Net Limited (2017)
dossier
that all related topics warrant further investigation. More-over, sufficient attention and support must be given to ex-ploring a detailed and sensible framework as well as oper-ational solutions to process, collaboration, and regulation questions. Only if all factors, perspectives and dependen-cies have been thought through, planned, and taken into account by all relevant stakeholders will digitalization and automatization be able to help the transportation, supply chain management and logistics sector achieve higher ef-ficiency built on trustful, secure, and motivated human perception.
References
Bauernhansl, T., ten Hompel, M., Vogel-Heuser, B. (2014). Industrie 4.0 in Produktion, Automatisierung und Logistik: Anwendung · Technologien · Migration.
Ruiner, C.; Klumpp, M. (2018a). 2018 Workshop at Florence School of Regulation (EUI), http://clubofflor-ence.org/?p=20, accessed 1 May 2018.
Ruiner, C.; Klumpp, M. (2018b). Results, http://clubof-florence.org/?p=61, accessed 1 May 2018.
Heistermann, F., Mallée, T., ten Hompel, M., Bundes-vereinigung Logistik (BVL) e. V. (Ed.) (2017). Digitalisie-rung in der Logistik, BVL Positionspapier. p. 9–15.
Kersten, W.; Seiter, M.; von See, B.; Hackius, N.; Maurer, T., Bundesvereinigung Logistik (BVL) e. V. (Ed.) (2017). Trends und Strategien in Logistik und Supply Chain Ma-nagement – Chancen der digitalen Transformation. p. 9–24.
Mesh-Net Limited (2017). What is the Internet of Things (IoT)?, https://www.mesh-net.co.uk/what-is-the-internet-of-things-iot/, accessed 26 April 2018.
Otto, B. et al., Fraunhofer-Gesellschaft e. V. (Ed.): Di-gitale Souveränität. Beitrag des Industrial Data Space. p. 8–11.
Figure 3. Comparison of risks and opportunities of digitalization Source: Kersten, W.; Seiter, M.; von See, B.; Hackius, N.; Maurer, T. (2017)
dossier
How to prepare workers for logistics innovations today and
tomorrow
Dominic Loske*
Artificial intelligence and the Internet of Things enable new and non-foreseeable potentials in various development and application scenarios. This disruptive and exponential development in informational technology has triggered a rapidly falling half-life of knowledge that underlines the importance of lifelong learning and the fast adaptation to new situations in logistics.
Introduction
T
he term logistics is ambiguous and has been exam-ined from different perspectives in scientific liter-ature. As an application-oriented scientific disci-pline, logistics analyzes the flow of goods in collaborative economic systems and provides recommendations for their design and implementation. Furthermore, logistics can be discussed as a branch that connects value chains of various dimensions (Pfeiffer, 2016). As an activity, logistics in-cludes the spatial-temporal transformation of goods, han-dling, packaging, order-picking, and sorting (Baumgarten et al., 2004).The aim of the present paper is to provide practice-orient-ed examples of applipractice-orient-ed procpractice-orient-edures when preparing workers for logistics innovations today and to discuss possible ways
for preparation, applied in the near future. Accordingly, the paper concentrates on logistics as an activity and pro-vides insights to the working fields of truck driving and transport planning.
Methodology
Practice-oriented examples are related to distribution logistics and provided by the expertise of a senior transport manager employed at one of the largest food retailing com-panies in Germany. After a brief introduction of a working system model, two cases are discussed. These cases concen-trate on a short work task description and an explanation of how workers are prepared for logistics innovations, both
in the past and today.
To be able to estimate how human workforce will be prepared in the future, it is necessary to include possible
Figure 1. Working system
dossier
developments in digital technologies. Therefore, a scenario technique used in scientific literature in order to estimate future development has been applied. The advantage of this method over other instruments is that the future is under-stood as a possibility space with many different directions of development and not as a fixed track (Gausemeier et al., 2009). The primary influencing factors are digitalization as a development boost causing disruptive changes and a shortage of specialists caused by the demographic change and an aging workforce (Eisenmann, Ittermann, 2017).
The working system as a frame of reference for prac-tice-oriented insights
In order to explain the way how workers are prepared for logistics innovations today, the model of a working system is used. The focus of the model is the work task, which is derived by a superior mission of the organization (Figure 1).
To accomplish the working task, the human resource impinges on the work object by using work equipment (Luczak, 1997). Successful interaction between humans
and the work equipment is necessary and requires a certain level of qualification. In addition to the input and output factors, exogenous factors such as the physical and social environment influence the working system and subsystems (Schultetus, 2006).
Case 1 – Preparing truck drivers for logistics innova-tions today and tomorrow
Truck drivers working in distribution logistics of food retail companies have to face changes in digital
innova-tions related to track-and-trace systems. Mobile devices or handheld scanners are currently used to display work tasks for the truck driver; for example, to load a certain amount of containers for a grocery store and deliver them within a given time window.
In order to fulfil such tasks, the truck driver must scan all relevant 1D barcodes, which are attached to the contain-ers, load them into his or her truck, and record differences between the data provided by the mobile device and the determined condition of transported goods. Furthermore, the mobile device is used to record the activities in reverse logistics. Figure 2 describes the working system of a truck driver in distribution logistics of food retail companies.
Currently, all truck drivers at a depot are trained inten-sively whenever disruptive innovations in track and trace systems take place. At the food retailing company exam-ined in this case, an intensive seminar-based training was held for all truck drivers when the first track and trace sys-tem was integrated in 2007.
A second wave took place in 2014 with intensive
semi-nar-based training supported by a digital mock-up with a demo tour. The need for this preparation was, on one hand, derived by a new design; on the other hand, a differ-ent operating principle significantly extends the processes covered in the whole working system.
In 2018, a third wave will take place in which design and hardware will change fundamentally. The truck driv-ers will be prepared to face these changes by conducting a self-study with a digital mock-up-based demo tour, an
Figure 2. Working system of a truck driver in distribution logistics
dossier
online video playable on a smartphone to explain the func-tions and features of the mobile device, and two weeks of support from experts placed in the area of the warehouse where the truck drivers are loading their trucks.
New truck drivers who begin work after these waves of disruptive innovations occur due to an increase of trans-portation capacity in a depot or due to fluctuation effects are not prepared by intensive seminars but by on-the-job training from existing drivers. In this context, a loss of knowledge is expected and can also be traced back to the fact that new releases cause slight changes.
The expectation can be conditionally proven by faults occurring in the loading and scanning process, which are measured in the amount of containers that not scanned properly and were therefore left back in the warehouse. All contents described are summarized in Figure 3.
In the future we expect increased fluctuation of truck drivers, as well as more frequent and extensive changes in digital technologies. Without proper preparation, modifi-cations in the technosphere and infosphere will reinforce the truck driver’s potential for faults while fulfilling work-ing tasks. Consequently, the future approach is a constant refresh and training of knowledge for all truck drivers with-out waiting for disruptive changes in hardware or software.
Case 2 – Preparing dispatchers for logistics innova-tions today and tomorrow
Dispatchers working in distribution logistics of food re-tail companies have to face changes in digital innovations
related to transport planning systems. The working system is used to plan tours with the aim of ensuring a punctu-al, complete and cost-efficient delivery of grocery shops. The bases for their work are orders placed by the shops and available resources as trucks ready for transportation. Figure 4 describes the working system of a dispatcher in distribution logistics of food retail companies.
The increasing amount of available information (1) caused by the technical development of logistic systems in the last years triggered a performance enhancement of transport planning systems (2). As a result, a constant im-provement in the level of competences for dispatchers is required (3).
Today dispatchers are schooled in a manner similar to truck drivers in a seminar concept that is held whenever disruptive changes for the working equipment take place. New employees are prepared by an intensive training on the job that lasts at least half a year. There is currently no preparation when add-ons for the planning software are introduced, except when release descriptions are provided. These documents contain all new features without filtering relevance or importance and are written from a technical point of view because they are provided by the IT business unit.
The result is an increasing amount of features that are not used despite being available. Figure 5 illustrates this by contrasting available and used features. The develop-ment of available functions is based on software releases and updates. The increasing usage of the software in
Q1-Figure 3. Relation of estimated training level and faults in loading/ scanning process for 2014 to 2017 Source: Authors’ own illustration
dossier
2016 and Q4-2016 is not attributable to the preparation of employees but to a change of the work task.
The scientific literature predicts an exchange of roles in logistic systems with a transfer of executive human work to
computer systems. Therefore, employers have to search for possibilities to transmit new available features to employ-ees and to lower possible acceptance and resistance hurdles when increasing automation or implementing artificial intelligence (Klumpp, 2017). This will be increasingly im-portant due to the fast, disruptive, and exponential devel-opment in informational technology and due to techno-logical advances. A rapidly falling half-life of knowledge underlines the importance of lifelong learning and the fast adoption to new situations in logistics (Wróbel-Lachows-ka, 2018).
Conclusion
This paper has presented practice-oriented examples re-lated to distribution logistics provided by the expertise of a senior transport manager employed at one of the largest food retailing companies in Germany in order to gain in-sights into how workers are prepared for logistics innova-tions today. As a frame of reference, the working system introduced by Hardenacke et al. was applied to the cases of how truck drivers and how dispatchers are prepared for logistics innovations today. Similarities of the two cases are the emergence of an aging workforce and a lack of spe-cialists, the preparation whenever disruptive innovations take place, and the type of training that is carried out by seminar-based training methods and recently with digital mock-ups.
To prepare truck drivers and dispatchers for logistics in-novations, a rapidly falling half-life of knowledge can be observed that underlines the importance of lifelong learn-ing and fast adaptation to new situations in logistics. Em-ployers have to turn away from inflexible training rhythms and start to adapt permanent methods for qualifying workforce. Schooling truck drivers through gamification, such as the MARTINA application provided by FOM University of Applied Science, while they are waiting for new working tasks can be understood as a role model of efficient, effective, and constant improvement of the qual-ification of workforce.
Figure 4. Working system of a dispatcher in distribution logistics
Source: Authors’ own illustration based on Hardenacke et al. (1985), in Luczak (1997)
Figure 5. Relation of available und used functions of transport planning system for 2014 to 2017
dossier
References
Baumgarten, H., Zadek, H., and Darkow, I.-L., eds. (2004), Supply Chain Steuerung und Services: Logistik-Di-enstleister managen globale Netzwerke - Best Practice, (Hei-delberg, Berlin: Springer).
Eisenmann, M., and Ittermann, P. (2017), ‘Hybride Dienstleistungen und Wandel der Arbeit: Herausforde-rungen und Perspektiven in der Logistik’, Soziologisches Arbeitspapier, 50: 1–46.
Gausemeier, J., Plass, C., and Wenzelmann, C., eds. (2009): Zukunftsorientierte Unternehmensgestaltung: Stra-tegien, Geschäftsprozesse und IT-Systeme für die Produktion von morgen, (Munich: Hanser).
Klumpp, M. (2017), ‘Automation and artificial intelli-gence in business logistics systems: Human reactions and collaboration requirements’, International Journal of Logis-tics Research and Applications: 1–19.
Luczak, H. (1997), ‘Kerndefinition und Systematiken der Arbeitswissenschaft’ in: H. Luczak, et al., Handbuch Arbeitswissenschaft (Stuttgart: Schäffer-Poeschel).
Luczak, H., Volpert, W., and Müller, T., eds. (1997): Handbuch Arbeitswissenschaft, (Stuttgart: Schäffer-Poe-schel).
Pfeiffer, S. (2016), ‘Bildung und Intralogistik in der In-dustrie 4.0 – eine empirische Annäherung’, Arbeit, 25 (3): 195–215.
Schultetus, W., eds. (2006), Arbeitswissenschaft von der Theorie zur Praxis: Arbeitswissenschaftliche Erkenntnisse und ihr wirtschaftlicher Nutzen, (Cologne: Köln: Wirtschafts-verlag Bachem).
Wróbel-Lachowska, M., Wisniewski, Z., and Polak-Sop-ińska, A. (2018), ‘The role of the lifelong learning in Logis-tics 4.0’, International Conference on Applied Human Fac-tors and Ergonomics, 1: 402–409.
dossier
The impact of digital technologies and artificial intelligence
on production systems in today Industry 4.0 environment
Francesco Pilati and Alberto Regattieri*
The industrial environment is currently experiencing its fourth industrial revolution, distinguished by the ubiquitous use of sensors that are able to capture large volumes of data regarding production processes. This vast quantity of digital information represents the raw material of the 21st century, which is able to fuel the decision processes of the factories of the future. The development and exploitation of novel algorithms and methods derived from cognitive processes of human beings represents the latest trend, both for research and application in the industrial sector. The adoption of artificial intelligence tools and techniques to design and manage smart assembly and manufacturing systems is the core of this manuscript. Two real industrial applications are presented to test and validate the afore-described approach, analysing both the advantages and the drawbacks of such solutions. In particular, a hardware/software architecture based on depth cameras is developed to digitalize the operator motions within assembly or manufacturing systems, whereas a set of neural algorithms defines the maintenance policies for continuously condition-monitored production systems.
Introduction
T
he industrial environment is currently experienc-ing what has been described as the fourth indus-trial revolution, namely Industry 4.0 (I40). The ubiquitous usage of sensors, which communicate through a world-wide network, make it possible to connect, in re-al-time, several entities of production systems, such as ma-chinery, equipment, final products, components, workers, suppliers, customers, etc. Together, these elements com-prise the Internet of things (IoT) (Stankovic, 2014). The huge volume of data produced by these connect objects is the raw material of the 21st century (Bortolini et al., 2017).These technologies facilitate the development of a new production paradigm, which has been termed personalized production. This paradigm satisfies the customer’s contem-porary need to participate in the production process since the product design phase (Hu et al., 2011).
Furthermore, today’s industrial environment is distin-guished by three trends of extreme importance. First, the workforce is aging alarmingly. In the last three lustrums, the percentage of European employees older than 50 years increased by 10 percent; that is, from 20 percent to 30 percent of the total working population (OECD, 2015). Currently, 5.8 million European workers are 60 years or older (7.4 percent of the entire workforce). Second, West-ern countries, Europe in particular, are experiencing the re-shoring of production plants that were previously off-shored to emerging countries (Ellram et al., 2013). This important trend is driven by growing labour costs in emerging countries that are almost equal to those in West-ern countries; the higher soft and digital skills of WestWest-ern workers compared to those of emerging country
competi-tors; the remarkable savings in transportation costs, which benefits local supply chains; as well as the flexibility and responsiveness needed to meet customer expectations (Davies, 2015).
Such a digital industrial environment generates a huge volume of date at high pace. Therefore, the development
and adoption of appropriate models and methods, as well as algorithms and techniques, is of major importance to obtain meaningful information from these data sets. One of the latest trends is represented by the adoption of biol-ogy-inspired algorithm, as artificial intelligence (AI). The definition of AI can be provided considering two relevant
* Francesco Pilati and Alberto Regattieri, Department of Industrial Engineering – University of Bologna, Viale del Risorgimento 2, 40136 Bologna, Italy,
Figure 1. Artificial intelligence definitions proposed in the literature considering the thinking-acting and
human-ity-rationality dimensions Source: Russel and Norvig, 2003
dossier
dimensions: thinking vs. acting and humanity vs. ration-ality (Russel and Norvig, 2003). The former measures the degree of reasoning against behaviour in a decision process, whereas the latter compares the decision process with hu-man versus purely rational approaches. Considering these dimensions, Figure 1 proposes four relevant but different definitions of AI that have been suggested in the literature.
This technology ensures a precise measurement of the op-erator absolute positons in the industrial 3D environment in relation to the difference pieces of equipment, products, and furniture displaced in the shop floor area. The MAS automatically, quantitatively, and dynamically evaluates a set of key performance indicators (KPIs) that deal with the monitored production process both from an ergonomic and a logistic perspective.
The large volume of data acquired by the MOCAP sys-tem represents the absolute geometric coordinates of each joint of the operator’s body on the shop floor, and therefore his skeleton posture. The developed MAS leverages this in-formation from an ergonomic perspective to dynamically evaluate the angle of every human body articulation and the related movement over the monitored time (Figure 3a). These data are further processed by the MAS to dy-namically and automatically assess several ergonomic indi-ces that evaluate the postures and movements of operators during working activities. For instance, the OWAS, REBA, NIOSH and EAWS can be easily evaluated by leveraging the distinctive features of the MAS. A useful tool provid-ed by the MAS is the automatic analysis of index trends in relation to the risk categories and specific body parts,
Figure 3a. Body articula-tion angles, knee
exempli-fication.
Figure 3b. Ergonomic index trend over time and per body part.
Source: Author’s own compilation
Figure 2a. Configuration of the MAS network
of depth cameras Figure 2b. Skeleton joints of the acquired human body Source: Author’s own compilation
dossier
which makes it possible to identify those tasks or manual activities that require corrective actions (Figure 3b).
Concerning the productive performances of the moni-tored operator within the shop floor, the MAS automati-cally and quantitatively evaluates the following set of KPIs: travelled distance and velocity of the operator and his dif-ferent body parts; vertical movements due to lifting and lowering activities; worker’s travelled paths and trajectories of his hands (Figure 4a); picking activity in-depth assess-ment (visited locations, duration, frequency, etc.) (Figure 4b); and working time partitioning, such as the distinction between added-value (task execution) and no added-value (walking, picking, etc.) activities.
Data analytics for condition based maintenance
A remarkable opportunity in the field of production sys-tem maintenance is represented by the adoption of data analytics tools and techniques that exploit AI algorithms, in particular for continuous condition monitoring. The aim of the maintenance policies is to define the optimal instant to perform a maintenance intervention, whether a component repair or replacement, in order to minimize the production system total breakdowns and maximize its techno-economic performance. A recent trend in this field of research is represented by the definition of the mainte-nance policies through the continuous monitoring of one or more relevant operating parameters of the considered production system. Thus, the maintenance interventions are defined considering the real-time conditions of one or more continuously measured parameters. Appropriate AI algorithms have to be developed and trained to define which alert threshold of the monitored parameter requests
an immediate maintenance intervention. A real industrial application of the proposed approach is represented by a fresh pasta production system. A particular system com-ponent determined several severe breakdowns in the entire production process. A customized AI algorithm has been developed to determine which operating parameter to monitor, along with its value, which distinguishes between a safe, warning and breakdown working zone. The follow-ing Figure 5 presents this exemplification.
The approach describe above is distinguished by several opportunities that positively affect the technical and eco-nomic performances of the analyzed production system during its entire lifetime. In particular, the most relevant
advantages of maintenance policies based on the continu-ous condition monitoring of an operating parameter are listed below (Alsina et al., 2018):
• Enhanced plant, production system, or machine availability
• Lower total cost of ownership of the considered pro-duction system
• Improvement in the design of complex production systems
• Potential revenue source for the maintenance depart-ment due to the sale of added-value services such as RAM analysis, maintenance strategy optimization, and forecasting of spare-part consumption.
Beside these positive aspects, the adoption of AI algo-rithm for the definition of maintenance policies, in par-ticular the exploiting of continuous condition monitor-ing, is distinguished by some possible but severe threats.
Figure 4a. Travelled paths and trajectories of operator and his
hands. Figure 4bpicking from shelves.. Assessment of component Source: Author’s own compilation
dossier
First, the determination of the link between a weak mon-itored signal and the component or system reliability is significantly challenging. The most relevant decisions deal with the definition of which parameter to monitor, which is the link between the parameter value and the time before failure and whether or not to link the condition information with a known failure state of the monitored component. Furthermore, a major challenge is represented by the definition of a proper measurement chain and the storage of the collected data. The relevant decisions deal with the identification of which sensor is appropriate for the monitored process, where to place the sensor consider-ing the appropriate necessary space, the storage capacity of the collected data with a demanding amount of data to be managed, and customer unwillingness to provide the data about their production processes. Finally, the first signals to be used, at low cost, are those typically needed to con-trol the technological process of machines; these are often available but neglected, such as positions, speed, currents, and temperatures.
The application of different AI algorithms to several real industrial production processes to define the optimal maintenance policy confirms this conflicting trend. Some of the approaches that have been tested in various case studies, with mixed results, are listed below:
• An artificial neural network was developed and adopted to forecast the spare-part consumption of packaging machinery, with very positive results. • A support vector machine was adopted to forecast
the reliability of mechanical and electric components in a refrigerating plant, with positive results along with some potential threats.
• A random forest algorithm was implemented to re-mote monitor a cutting machine reliability assessing the different occurring alarms. The difficulty of
Figure 5. Maintenance policy based on the continuous condition monitoring of an operating parameter. Source: Author’s own compilation
assessing and subsequently exploiting the different clusters of alarms led to negative results.
Conclusion and further research
This paper proposes the adoption of AI tools and tech-niques to design and manage production systems in the Industry 4.0 environment. Two real industrial applications have been presented to test and validate the approach de-scribed above, analysing both the advantages and also the drawbacks that distinguish such solutions. In particular, a hardware/software architecture based on depth cameras was developed to digitalize the operator motions within assembly or manufacturing systems, whereas a set of neural algorithms defines the maintenance policies for continu-ously condition monitored production systems.
The main outcomes of this research suggest that current and future technological resources offer interesting oppor-tunities to exploit AI tools and algorithms for production systems. AI can accelerate the development of strategies to monitor these systems, with a particular focus on human performances and to define proper and efficient mainte-nance approaches. A relevant risk is represented by the fact that this new paradigm could eventually emphasize the current problems related to data collection and interpreta-tion. Unfortunately, from an engineering perspective, it is not foreseeable to significantly reduce the difficulty deter-mined by the monitored data interpretation. Thus, further research should focus the effort to reinforce the process to analyze the huge quantity of collected data, along with providing meaningful information that has strong rela-tions and backgrounds to real industrial applicarela-tions.
dossier
References
Alsina, E. F., Chica, M., Trawiński, K. and Regattieri, A. (2018). On the use of machine learning methods to pre-dict component reliability from data-driven industrial case studies. The International Journal of Advanced Manufac-turing Technology, 94(5–8), 2419–2433.
Atzori, L., Iera, A., and Morabito, G. (2010). The In-ternet of Things: A survey. Computer Networks, 54(15), 2787–2805.
Azuma, R., Baillot, Y., Behringer, R., Feiner, S., Julier, S., and MacIntyre, B. (2001). Recent advances in augmented reality. IEEE Computer Graphics and Applications, 21(6), 34–47.
Bellman, R. E. (1978). An Introduction to Artificial In-telligence: Can Computers Think? Boyd & Fraser Publi-shing Company.
Bortolini, M., Ferrari, E., Gamberi, M., Pilati, F. and Fac-cio, M. (2017). Assembly system design in the Industry 4.0 era: a general framework. IFAC-PapersOnLine, 50(1), 5700–5705.
Bortolini, M., Gamberi, M., Pilati, F. and Regattieri, A. (2018). Automatic assessment of the ergonomic risk for manual manufacturing and assembly activities through op-tical motion capture technology. Procedia CIRP, in press.
Charniak, E. and McDermott, D. (1985). Introduction to Artificial Intelligence. Addison-Wesley.
Davies, R. (2015). Industry 4.0. Digitalisation for productivity and growth. European Parliamentary Re-search Service. Retrieved from http://www.europarl.eu-ropa.eu/thinktank/it/document.html?reference=EPRS_ BRI(2015)568337.
Ellram, L. M., Tate, W. L., and Petersen, K. J. (2013). Offshoring and reshoring: an update on the manufactur-ing location decision. Journal of Supply Chain Manage-ment, 49(2), 14–22.
Flach, P. (2012). Machine learning: the art and science of algorithms that make sense of data. Cambridge University Press.
Gantz, J., and Reinsel, D. (2011). Extracting value from chaos. IDC iview, Framingham.
Hu, S. J., Ko, J., Weyand, L., Elmaraghy, H. A., Lien, T. K., Koren, Y., … Shpitalni, M. (2011). Assembly system design and operations for product variety. CIRP Annals – Manufacturing Technology, 60(2), 715–733.
Kurzweil, R. (1990). The Age of Intelligent Machines. MIT Press.
OECD (2015). OECD Employment Outlook 2015, OECD Publishing, Paris.
Poole, D., Mackworth, A. K., and Goebel, R. (1998). Computational intelligence: A logical approach. Oxford University Press.
Russel, S. and Norvig, P. Artificial intelligence: A modern approach, 2003. EUA: Prentice Hall, 178.
Stankovic, J. (2014). Research directions for the internet of things. IEEE Internet of Things Journal, 1(1), 3–9.
dossier
Liability and automation: legal issues in autonomous cars
Giuseppe Contissa
a, Francesca Lagioia
a,b, Giovanni Sartor
a,bThe deployment of highly automated systems, such as autonomous cars, is going through an accelerated expansion: as usual for emerging disruptive technologies a slow start is followed by a more and more rapid surge. One of the most important legal issues concerning these systems is related to liability for accidents. In particular, highly automated systems will make choices and engage in actions – usually with some level of human supervision, or even without any such supervision. In this context, there is the need to analyse how the decision-making process is split between humans and machines, and critically revise the way tasks, roles, and liabilities are allocated. In this contribution we analyse the impact of automation in the allocation of liability within autonomous cars. We first discuss the tasks allocation between human and automation, and the resulting responsibilities. Then, we analyse how the introduction of different levels automation gives rise to a redistribution of tasks between human and automation and, therefore, a reallocation of the liability burden between the user and the manufacturer.
Task-responsibilities and the impact of automation
I
n order to introduce the analysis of liability issues, we need to refer to the concept of task-responsibility, i.e. the duty pertaining to the correct performance of a certain task or role.First of all, we need to identify task-responsibilities of the user, since their violation may result in personal liability. In fact, whenever there is a failure in a complex system, we try to connect the failure with the missing or inadequate execution of a task, and so with the (natural or legal) per-sons who were responsible for that task. As a consequence of the failure to comply with their task-responsibilities, these persons are subject to blame, penalties, and/or the payment of damages.
Secondly, we need to identify task-responsibilities of the automated system, namely, the requirements the sys-tem should comply with. These are also relevant, since a failure to meet them may make the system’s producers or maintainers liable. With the introduction of higher lev-els automation, as task-responsibilities are progressively delegated to technology, liability for damages shifts from human operators to the organisations that designed and developed the technology, defined its context and uses, and are responsible for its deployment, integration, and maintenance.
In this context, it is necessary to adopt a systematic ap-proach to match the degree of automation to different responsibilities of users of automated systems at different levels as well as to the responsibilities of other actors in-volved (managers, producers, and maintainers)(Contissa et al. 2013).
To this end, we consider the Level Of Automation Tax-onomy (LOAT), developed by SESAR 16.5.1 (Save and Feuerberg, 2014) used to assess the levels of automation introduced by a new technology and to determine the cor-responding impacts on the division of tasks between hu-mans and machines.
The LOAT table provides criteria for assigning a level of automation to a technology with regard to four differ-ent cognitive functions: information acquisition (A), in-formation analysis (B), decision-making (C), and action implementation (D). Figure 1 shows a simplified version of the LOAT: all columns start with level 0, correspond-ing to a fully manual accomplishment of the task, without any technical support. At Level 1 the task is accomplished with “primitive” technical tools, i.e., low-tech non-digital artefacts. From level 2 on upwards, “real” automation is involved, and the role of the machine becomes increasingly significant up to the level where the task is fully automated.
A certain technology may have different levels of automa-tion, according to whether the actors are dealing with the four cognitive functions mentioned above.
The LOAT expresses varying levels of interaction between humans and the technology in question. In the first in-stance, it is used to better understand technology. It pro-vides an accurate account of human-machine interaction and serves as a tool for refining the concept of automation. By conceptualizing automation on the basis of the human factor, it generates awareness of human-machine interac-tion.
